Lessons in Spatial Analysis

Lauren Rosenshein Bennett gives a crash course in six techniques for spatial analysis.

Jul 23rd, 2012

Start From:
Player Color:

Right-click on these links to download and save this video.


00:01So we're going to hear more from them later.

00:02But what all the students at Washington Lee share is a passion for solving spatial problems.

00:08Their class is about problem-based learning.

00:11So what we want to do is continue with that thought and provide a little more information...

00:15...about how we do spatial problem analysis and solve those kinds of problems.

00:20To do that, please welcome Lauren Rosenshein Bennett.

00:28Thanks, John. Instead of showing you one or two new tools and a specific application of those tools...

00:35...we're going to try something a little bit different this year, a kind of crash course in spatial analysis.

00:42We'll go through six techniques, from fundamentals like overlays to more advanced methods like interpolation.

00:49That way, you'll be ready to approach every problem with the best spatial analysis that ArcGIS 10.1 has to offer.

00:57The first technique is visual analysis, using geography to integrate multiple layers and gain new insight.

01:04In this example, we're looking at mortality rates 30 days after heart failure for Medicare beneficiaries.

01:14Now, ideally, we would expect the level of service provided to be consistent across the country...

01:20...but as we can see, there are areas of the country where those mortality rates are higher.

01:26So what could be going on here? Maybe it has something to do with hospital access.

01:32We can search for hospitals, add them to the map, and very quickly...

01:37...we can start to use a visual analysis to explore patterns and explore our data, all from ArcGIS Online.

01:46Now let's switch gears to technique number two, which is about overlays...

01:51...overlays that help us quantify the intersection of multiple datasets.

01:56In this example, we want to know the acreage of each type of land cover...

02:00...in Davy Crockett National Forest just south of Dallas, Texas.

02:05Prior to 10.1, to do this, we would've started by running an intersect, then follow that with a summary statistics...

02:14...and finally, we'd be able to compute - to create a Pivot table.

02:20We've combined this very common workflow into one simple tool called Tabulate Intersection.

02:28Now we have the acreage and the percentage of each type of land cover in the national forest...

02:34...all using an overlay that helps us compute the overlapping polygons and the intersection of those overlapping polygons.

02:43Technique number three is all about interpolating surfaces from discrete point data.

02:48In this example, we have data from USGS on lead concentrations in the state of Texas.

02:55So how do we create an accurate, continuous surface from this kind of sample data?

03:02In 10.1, a new tool, called empirical Bayesian kriging, does this for us with the simplicity of our most basic interpolation tools...

03:10...and the sophistication of our most advanced geostatistical methods.

03:16This analysis looks really simple, and it is really simple to run, but this is a lot more than just an interpolated surface.

03:25Because we're using a statistical approach to interpolation, we can look at not only the prediction surface...

03:31...but also the prediction error, giving us an understanding of the uncertainty in our predictions.

03:38These darker areas have a higher prediction error...

03:41...which is an indication that we may need to take additional samples in those locations.

03:47Which takes us to technique three and a half, which is all about spatial analysis everywhere.

03:53So let's grab our iPad and head out to the field just north of Austin.

03:58We'll start by collecting some samples.

04:00We'll add some data based on our latest soil samples, and when we're done, click this button.

04:11Sends that feature service to a geoprocessing service, which is running that advanced statistical interpolation, all from our iPad.

04:19It'll then return to us the latest surface based on that live data we're collecting out in the field.

04:31It really will.

04:43Okay. Not only do we get the latest surface based on the data we've collected out in the field...

04:52...we can also incorporate additional geoprocessing services.

04:56Oh, there we go. We can see pretty quickly how different that surface is based on that live data.

05:02And then we can run another service which is associated with this application, which is a watershed service.

05:09And what that will do is find the watershed that we're in as well as the average lead concentration in our watershed...

05:15...as well as the state of Texas and the rest of the United States.

05:19So by integrating these services throughout the platform, across devices, we're enabling everybody to do spatial analysis.

05:29Technique number four is network analysis, which is a type of analysis that many of you are doing every day.

05:36This technique is a little bit different because it's actually a sneak peek...

05:40...at the new ArcGIS Online world routing service being released this fall.

05:46We have a set of stops that we need to make this afternoon in the DC metro area...

05:50...and we want to route between them so that we can ensure we save time and money.

05:55So we'll use the Find Route tool, which we've hooked up to the new world routing service.

06:00The service combines advanced network analysis with worldwide optimized data, including live and historic traffic data.

06:11Now we can choose the time of day that we want to do the route, choose to optimize...

06:16...and send that analysis up to ArcGIS Online, up to that world routing service where it'll just do all the work for us.

06:23And like I said, this is a global service.

06:27With historic traffic data in 97 countries and live traffic data in 26 countries...

06:35...routing throughout Europe using live and historic traffic data is as simple as just a few clicks.

06:43Technique number five is about site selection, using demographics and market potential variables...

06:49...to choose the best location for a new store, for a targeted marketing campaign, or for a new public facility.

06:58In this example, we have a store that's performing really well in the Denver area...

07:03...and we have nine potential locations to put our next store.

07:07So how do we decide which is the best location?

07:11Well, we already know what success looks like, and we know the characteristics of our successful store.

07:17The powerful predictive tool in Business Analyst uses a principal components analysis...

07:22...to compare each of those potential locations to our successful store.

07:28We can choose from thousands of available variables, or we can choose an existing list, like the Retail Market Potential list...

07:37...which has close to 300 variables on consumer spending, census demographics, and more.

07:43The tool then uses a principal components analysis to assess the statistical similarity of each of those potential locations...

07:50...to our successful store and returns those ringed trade areas.

07:54So we can see the ones that are most similar to our successful store in darkest green.

08:00Taking the guesswork out of this kind of site selection can help us make more informed decisions.

08:07Technique number six is about finding clusters in our data.

08:12This takes us back to the Medicare data that we looked at earlier in ArcGIS Online.

08:17Now, we already saw that there's regional variation in these mortality rates, but we also know...

08:22...that assessing the quality and the cost of the Medicare system involves a lot more than just one variable.

08:28So we have data on readmission rates and standardized costs. So how do we make sense of all this data?

08:37A new tool in 10.1, called Grouping Analysis, lets us take these variables, bring them all together...

08:43...and create groups based on these variables that have similar characteristics.

08:50The first thing that we see is actually a spatial pattern based on those variables.

08:56But this report, which is also created by the Grouping Analysis tool, lets us dig in a little bit deeper.

09:01We can see that the red area in the Southeast is average in terms of mortality and readmissions...

09:07...but significantly higher in terms of spending than the rest of the country.

09:12So average level of service, much higher spending.

09:17This is an area we might want to focus our attention if our goal is cutting costs.

09:23The last thing that I want to show you is how easy it is for us to share our analysis using the new geoprocessing package.

09:30And this applies to techniques one through six and thousands more.

09:35Now, sharing our analysis is as simple as right-clicking on a result...

09:40...whether it's the simple run of a single tool or a complex model, and choosing to share it as a geoprocessing package.

09:47When we hit Share, it's taking our input data; our tools, which could include models with nested models and scripts...

09:55...any associated content; and our results, putting them all into one geoprocessing package...

10:01...and, in this case, sending them up to ArcGIS Online.

10:07Now we can take a look at our ArcGIS Online account and see that that geoprocessing package has been added.

10:14And this is a lot more than just the results of our analysis.

10:17This is our methodology, this is the tradecraft associated with that analysis.

10:22So we can share this with relevant groups and really contribute to the conversation in a meaningful way.

10:28These are just six of the thousands of ways that we can use spatial analysis to solve problems...

10:34...to make more informed, data-driven decisions.

10:38And you'll see all morning the way that spatial analysis plays a critical role in almost every workflow...

10:45...from the simple to the complex, from 3D to imagery, and everything in between. Thanks, John.

10:59Thanks, Lauren. One of the things that you might not have noticed is we're not talking about Desktop and Server...

11:05...and Mobile and Online, but instead, we're talking about one complete system.

11:10And spatial analysis, like everything else, cuts across the entire platform. All the boundaries are starting to blur together.

Copyright 2016 Esri
Auto Scroll (on)Enable or disable the automatic scrolling of the transcript text when the video is playing. You can save this option if you login


No comments. Be the first to write one below.

Comment on this Video